University of Groningen
An economic assessment of high-dose influenza vaccine
van Aalst, Robertus
DOI:
10.33612/diss.127973664
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
van Aalst, R. (2020). An economic assessment of high-dose influenza vaccine: Estimating the vaccine-preventable burden of disease in the United States using real-world data. University of Groningen. https://doi.org/10.33612/diss.127973664
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
3
Economic assessment of a high-dose versus a standard-dose influenza vaccine in the US Veteran population: estimating the impact on
hospitalization cost for cardio-respiratory disease
Robertus van Aalst a,b, Ellyn M. Russo c, Nabin Neupane c, Salaheddin M. Mahmud d,e, Vincent Mor f,g, Jan Wilschut h, Ayman Chit b,i, Maarten Postma a,j,k, Yinong Young-Xu c,l
a Department of Health Sciences, University Medical Center Groningen, University of Groningen,
Groningen, the Netherlands
b Regional Epidemiology and Health Economics, Sanofi Pasteur, Swiftwater, Pennsylvania, USA c Clinical Epidemiology Program, Veterans Affairs Medical Center, White River Junction, Vermont, USA d Department of Community Health Sciences, College of Medicine, University of Manitoba, Winnipeg, MB,
Canada
e George & Fay Yee Center for Healthcare Innovation, University of Manitoba/Winnipeg Regional Health
Authority, Winnipeg, MB, Canada
f Brown University, School of Public Health, Dept. Health Services, Policy and Practice, Providence, RI, USA g Providence VA Medical Center, Center of Long-Term Services and Support, Providence, RI, USA
h Department of Medical Microbiology, University Medical Center Groningen, University of Groningen,
Groningen, the Netherlands
i Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada
j Unit of PharmacoTherapy, -Epidemiology & -Economics (PTE2), University of Groningen, Department of
Pharmacy, Groningen, the Netherlands
k Department of Economics, Econometrics & Finance, University of Groningen, Faculty of Economics &
Business, Groningen, the Netherlands
l Department of Psychiatry, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA
Published on 26 July 2019 Vaccine. 2019;37(32):4499-503.
84 Chapter 3
ABSTRACT
ObjectiveTo compare the economic impact of high-dose trivalent (HD) versus standard-dose trivalent (SD) influenza vaccination on direct medical costs for cardio-respiratory hospitalizations in adults aged 65 years or older enrolled in the United States (US) Veteran’s Health Administration (VHA).
Methods
Leveraging a relative vaccine effectiveness study of HD versus SD over five respiratory seasons (2010/11 through 2014/15), we collected cost data for healthcare provided to the same study population both at VHA and through Medicare services. Our economic assessment compared the costs of vaccination and hospital care for patients experiencing acute cardio-vascular or respiratory illness.
Results
We analyzed 3.5 million SD and 158,636 HD person-seasons. The average cost of HD and SD vaccination was $23.48 (95% CI: $21.29 - $25.85) and $12.21 (95% CI: $11.49 - $13.00) per recipient, respectively, while the hospitalization rates for cardio-respiratory disease in HD and SD recipients were 0.114 (95% CI: 0.108 - 0.121) and 0.132 (95% CI: 0.132 - 0.133) per person-season, respectively. Attributing the average cost per hospitalization of $11,796 (95% CI: $11,685 - $11,907) to the difference in hospitalization rates, we estimated savings attributable to HD to be $202 (95% CI: $115 – $280) per vaccinated recipient.
Conclusions
For the five-season period of 2010/11 through 2014/15, HD influenza vaccination was associated with net cost savings due to fewer hospitalizations, and therefore lower direct medical costs, for cardio-respiratory disease as compared to SD influenza vaccination in the senior US VHA population.
85
BACKGROUND
Adults 65 years and older (hereinafter referred to as seniors) are at an increased risk for complications caused or triggered by an influenza infection [1]. Young-Xu and colleagues estimated the range of annual direct medical costs of influenza-attributable hospitalizations at Department of Veterans Affairs (VA) Medical Centers for senior Veterans Health Administration (VHA) enrollees over five respiratory seasons (2010/11 through 2014/15) to be between 24 and 34 million US dollars [2]. Given this substantial cost, a health economic analysis of the various influenza vaccination strategies for this age group is pertinent.
One of the vaccination options available to the VHA during this period was the injectable high-dose inactivated trivalent influenza vaccine (Fluzone® High-Dose, Sanofi Pasteur, PA, US, licensed in the US in 2009 for people aged 65 years and older; hereinafter referred to as the high-dose vaccine (HD)). HD contains four times more influenza hemagglutinin antigen than standard-dose trivalent influenza (SD) vaccines (60 μg vs. 15 μg per strain), improving immune response and therefore protection, in seniors. Young-Xu et al. [3] reported a relative vaccine effectiveness (rVE), or additional reduction, of HD versus SD of 10% (95% CI, 8% –12%) for all-cause hospitalization; 18% (95% CI, 15% – 21%) for cardio-respiratory-associated hospitalization; and 14% (95% CI, 6% –22%) for influenza/pneumonia-associated hospitalizations during five respiratory seasons (2010/11 through 2014/15).
Various features of the Young-Xu et al. (2019) study enable us to assess the contribution of HD in lowering the direct medical cost of influenza-attributable hospitalizations, thereby improving the accuracy of economic burden estimations. First, the study included hospitalizations in non-VA medical centers. The majority of senior Veterans are “dual users” and receive care in both VA and non-VA facilities paid for by Medicare [4, 5]. Second, the study captured five rather than one single respiratory season. Incorporating seasonal variation in influenza viral circulation and vaccine effectiveness increases the confidence in our economic assessment as an average economic effect. Third, the study used a statistical method to adjust for observable and unobservable differences between the HD and SD recipients.
86 Chapter 3
In this paper, we will assess the economic impact of HD versus SD vaccination on cost of hospitalization for cardio-respiratory disease in the population analyzed by Young-Xu et al. (2019). In addition, we estimate the economic impact in a scenario where 10% of the study population had received HD and 90% SD.
METHODS
Study Design, Population and Data Sources
The Young-Xu et al. [3] study, a retrospective cohort study with approximately 700,000 patients included in each of the five respiratory seasons, compared hospitalizations between those who received HD versus SD at a VA facility. Patients were included when they were at least 65 years old at vaccination, had received only one HD or SD vaccine in the seasons of interest, and had sought medical care at a VA facility in the six months before vaccination. This resulted in a study population of 3.5 million SD and 158,636 HD person-seasons. We used the same population and methods of Young-Xu et al. (2019) to calculate rVEs for the present study. In summary, for each study participant at each season, the baseline period (during which baseline characteristics were measured) was defined from the beginning of each respiratory season in week 27 (beginning of July) until his or her influenza vaccination date. The observation period (during which study outcomes were measured) was defined from two weeks after vaccination until the end of the respiratory season in week 26 (end of June). Crude rVE rates were adjusted for treatment selection bias (confounding by indication) using differences in observable baseline characteristics between the cohorts that included demographics, comorbidities adapted from the Deyo-Charlson comorbidity score [6], and VA priority group, a surrogate measure for socio-economic status (Appendix 1) [7]. In addition, an instrumental variable (IV) based on the facility’s treatment preference for HD, defined as the proportion of HD recipients at a certain facility in a given respiratory season, was used to act as a pseudo-randomizer of unobservable differences [3].
VHA is the largest integrated health care system in the US, providing care at 1,240 health care facilities, including 170 VA Medical Centers and 1,061 outpatient sites of care of varying complexity (VA outpatient clinics) to over nine million Veterans enrolled in healthcare through VA [8]. Admissions to VA hospitals were derived from
87 its unified electronic medical record system (EMR) that contains information about inpatient, outpatient, and emergency department (ED) visits.
For the cost of vaccination in VA facilities, we obtained data from the National Acquisition Center Contract Catalog Search Tool [9]. Hospitalizations, and their reimbursement costs, of VHA enrollees that occurred in non-VA facilities were obtained from the Centers for Medicare and Medicaid Services (CMS) administrative fee-for-service claims. These records supplement those in the VHA database as many patients seek healthcare outside VA once eligible for CMS benefits. While VHA applies a system of cost allocation, costs of non-VA hospitalizations are based on insurance reimbursements, which do not necessary reflect true costs for the healthcare provider [10].
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes
96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑁𝑁𝑁𝑁𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3)
3
88 Chapter 3
adjusted for number of HD and SD recipients (
96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑌𝑌𝑌𝑌�𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3) and 96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑌𝑌𝑌𝑌�𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3) , respectively), to estimate
the number of outcomes in the SD cohort (2).
96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑌𝑌𝑌𝑌�𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3) (1) 96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑌𝑌𝑌𝑌�𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3) (2) 96
The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our primary outcome of interest was an acute hospitalization for cardio-respiratory disease, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, ICD-9: 390-519, Appendix 2). Because this definition is less inclusive than the definition by Young-Xu et al. (2019) that included both acute hospitalizations and nursing home admissions, we recalculated rVEs using the same statistical model for cardio-respiratory disease. We assessed HD’s impact on cost of acute hospitalization for pneumonia or influenza (P&I) as well. Because expected underreporting of these hospitalizations introduced significant bias in the estimation of incidence rates, we refer for the results to appendices 3 and 4 [11-13]. In addition, we report HD’s impact on a more sensitive outcome, all-cause hospitalizations, in these appendices.
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We used a model based on the incidence rate ratio (RR) derived from the IV-adjusted rVE (1) and applied the RR to the total number of outcomes (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇), adjusted for number of HD and SD
recipients (𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 and 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻, respectively), to estimate the number of outcomes in the SD cohort (2).
𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = (1 − 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅) × 100% ; 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 =𝑌𝑌𝑌𝑌𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻�𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 � (1) 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=1+𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (2) Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻=𝑌𝑌𝑌𝑌�𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 (3) (3) Where 97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
is the study population size in a given respiratory season and consisting of all vaccine recipients (
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
);
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population;
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
and
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
are the number of outcomes attributed
to the SD and HD recipients
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
is the IV-adjusted incidence rate ratio; and
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
is the hospitalization rate of outcome
97
Where 𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 is the study population size in a given respiratory season and consisting of all vaccine recipients (𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇= 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻+ 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻); 𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇 is the total number of observed outcomes (e.g. hospitalizations for cardio-respiratory disease) in the study population; 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻and 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 are the number of outcomes attributed to the SD and HD recipients (𝑌𝑌𝑌𝑌𝑇𝑇𝑇𝑇= 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻+ 𝑌𝑌𝑌𝑌�𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻); 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 is the IV-adjusted incidence rate ratio; and
Hospitalization rate𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the hospitalization rate of outcome 𝑌𝑌𝑌𝑌�𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one
hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the cost-savings per patient. The maximum (potential) savings
in the group of SD recipients.
After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the SD cohort. The multiplicative inverse of ARR results in the number needed to treat (NNT = 1/ARR): the number of patients that need to be switched over from SD to HD to prevent one hospitalization. For cost of vaccination, we averaged season specific vaccine and administration costs. To increase the accuracy of the economic assessment and reduce the impact of data entry errors and rare-but-extreme values, we removed the top and bottom two percentiles equivalent to at least two standard deviations in a normal distribution of the observed hospitalization costs, as was done by Young-Xu et al. (2017), retaining 96% of observations in the cost-analysis [2, 14]. Because variation in vaccination cost was small, we included all observations. We used random sampling with replacement bootstrapping to calculate 95% confidence intervals (CI).
To evaluate cost-savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the number needed to treat (NNT) minus the average cost difference of administering the two vaccines. The cost of administering a vaccine included the cost of the vaccine itself as well as the cost of the administration process (vaccine injection and record keeping). We calculated the total realized cost-savings by multiplying the total number of HD recipients by the
89 cost-savings per patient. The maximum (potential) savings were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
(4) Where
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
is the number of observed outcomes in the SD cohort,
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
are the total costs of these outcomes,
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
the number of HD recipients,
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
the total cost of vaccinating the HD cohort,
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
the number of SD recipients, and
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
the total cost of vaccinating the SD cohort.
98
were calculated by dividing the total savings by the HD proportion minus the number of HD recipients divided by the sum of HD and SD recipients.
SavingsSD➝HD = 1 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑇𝑇𝑇𝑇 − � 1 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻− 1 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻∑ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻� (4)
Where 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻is the number of observed outcomes in the SD cohort, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑌𝑌𝑌𝑌𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 are the total costs of these outcomes, 𝑁𝑁𝑁𝑁𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the number of HD recipients, 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻 the total cost of vaccinating the HD cohort, 𝑁𝑁𝑁𝑁𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the number of SD recipients, and 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆𝑆𝑆𝐻𝐻𝐻𝐻 the total cost of vaccinating the SD cohort. SavingsSD➝HD are the
estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI.
We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings.
Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our
economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential
are the estimated savings per SD recipient if they had received an HD vaccination instead.
The lower limit of the CI for cost-savings is based on the lower limit of the CI for the rVE, the upper limit for incremental costs of vaccination, and the lower limit of hospitalization costs. We applied the opposite limits of the ones used to calculate the lower limit for the upper limit. The variation in VHA costs and Medicare reimbursements, as well as variation in the cost of vaccination, are reflected in their CI. We first calculated season-specific NNTs and cost-savings using season-specific numbers of HD and SD recipients, observed hospitalizations and costs, for which the results are presented in Appendices 3 and 4. We then analyzed combined data from all five seasons longitudinally, accounting for repeated measures from patients appearing in multiple seasons, to provide one summary measure of NNT and cost-savings. Sensitivity Analysis
We performed three sensitivity analyses to test the robustness of our findings. First, our economic assessment reassigns the observed total number of outcomes to the HD and SD cohorts using the IV-adjusted rVE. We assume that the true proportion of outcomes in VHA-hospitals does not change after this reassignment. To explore the sensitivity of the cost-savings to potential changes of the true proportion, we varied the observed proportion of hospitalizations with underlying cardio-respiratory disease by 25 percentage points in either direction. This allowed us to model a best/ worst case scenario without negative hospitalizations in any of the five respiratory seasons (Appendix 6). Second, we report a historical cost-assessment that will change if underlying costs change. To explore the sensitivity of the historical cost savings to changes in the incremental cost of vaccination – the average cost difference of